id
stringlengths
33
34
title
stringlengths
6
218
authors
stringlengths
5
6.98k
categories
stringlengths
5
188
summary
stringlengths
34
3.26k
link_pdf
stringlengths
33
34
updated
timestamp[s]
published
timestamp[s]
embeddings
dict
http://arxiv.org/abs/2306.01327v1
Speech Translation with Foundation Models and Optimal Transport: UPC at IWSLT23
Ioannis Tsiamas, Gerard I. Gállego, José A. R. Fonollosa, Marta R. Costa-jussà
cs.CL, cs.SD, eess.AS
This paper describes the submission of the UPC Machine Translation group to the IWSLT 2023 Offline Speech Translation task. Our Speech Translation systems utilize foundation models for speech (wav2vec 2.0) and text (mBART50). We incorporate a Siamese pretraining step of the speech and text encoders with CTC and Optimal...
http://arxiv.org/pdf/2306.01327v1
2023-06-02T07:48:37
2023-06-02T07:48:37
{ "title": [ -0.039489746, -0.013259888, -0.014533997, -0.041107178, -0.030914307, 0.0009813309, -0.048980713, 0.0037460327, -0.000995636, 0.05609131, 0.0037631989, -0.007396698, 0.016082764, -0.03817749, -0.02420044, 0.008255005, -0.010803223, 0...
http://arxiv.org/abs/2306.01325v1
LyricSIM: A novel Dataset and Benchmark for Similarity Detection in Spanish Song LyricS
Alejandro Benito-Santos, Adrián Ghajari, Pedro Hernández, Víctor Fresno, Salvador Ros, Elena González-Blanco
cs.CL, cs.IR
In this paper, we present a new dataset and benchmark tailored to the task of semantic similarity in song lyrics. Our dataset, originally consisting of 2775 pairs of Spanish songs, was annotated in a collective annotation experiment by 63 native annotators. After collecting and refining the data to ensure a high degree...
http://arxiv.org/pdf/2306.01325v1
2023-06-02T07:48:20
2023-06-02T07:48:20
{ "title": [ -0.022766113, -0.0041923523, -0.045135498, -0.045715332, 0.012580872, 0.00035595894, -0.03152466, 0.021896362, 0.019165039, 0.039855957, -0.046936035, 0.024154663, -0.008628845, 0.009384155, -0.06210327, -0.005844116, -0.019729614, 0...
http://arxiv.org/abs/2306.01318v1
Text Style Transfer Back-Translation
Daimeng Wei, Zhanglin Wu, Hengchao Shang, Zongyao Li, Minghan Wang, Jiaxin Guo, Xiaoyu Chen, Zhengzhe Yu, Hao Yang
cs.CL, cs.LG
Back Translation (BT) is widely used in the field of machine translation, as it has been proved effective for enhancing translation quality. However, BT mainly improves the translation of inputs that share a similar style (to be more specific, translation-like inputs), since the source side of BT data is machine-transl...
http://arxiv.org/pdf/2306.01318v1
2023-06-02T07:33:47
2023-06-02T07:33:47
{ "title": [ -0.019821167, -0.009750366, -0.019958496, -0.021331787, -0.049072266, -0.0061683655, -0.053131104, -0.012199402, -0.0062675476, 0.08691406, 0.01235199, 0.0016613007, 0.07623291, -0.041625977, 0.00030875206, 0.036224365, 0.036895752, ...
http://arxiv.org/abs/2306.01312v2
Syntax-aware Hybrid prompt model for Few-shot multi-modal sentiment analysis
Zikai Zhou, Haisong Feng, Baiyou Qiao, Gang Wu, Donghong Han
cs.CL
Multimodal Sentiment Analysis (MSA) has been a popular topic in natural language processing nowadays, at both sentence and aspect level. However, the existing approaches almost require large-size labeled datasets, which bring about large consumption of time and resources. Therefore, it is practical to explore the metho...
http://arxiv.org/pdf/2306.01312v2
2023-07-31T09:03:40
2023-06-02T07:22:43
{ "title": [ 0.0042152405, -0.013771057, -0.009719849, -0.049804688, -0.019760132, 0.007385254, -0.06112671, -0.0001436472, 0.040649414, 0.029251099, -0.042755127, -0.01109314, 0.019241333, -0.007762909, -0.07696533, -0.02229309, -0.0034618378, -...
http://arxiv.org/abs/2306.01311v1
MetaVL: Transferring In-Context Learning Ability From Language Models to Vision-Language Models
Masoud Monajatipoor, Liunian Harold Li, Mozhdeh Rouhsedaghat, Lin F. Yang, Kai-Wei Chang
cs.CL
Large-scale language models have shown the ability to adapt to a new task via conditioning on a few demonstrations (i.e., in-context learning). However, in the vision-language domain, most large-scale pre-trained vision-language (VL) models do not possess the ability to conduct in-context learning. How can we enable in...
http://arxiv.org/pdf/2306.01311v1
2023-06-02T07:21:03
2023-06-02T07:21:03
{ "title": [ -0.02243042, -0.017044067, -0.03894043, -0.031173706, -0.020385742, -0.0127334595, -0.020980835, -0.017684937, 0.008743286, 0.03451538, -0.027954102, -0.002286911, 0.010826111, -0.02319336, -0.036010742, -0.0074501038, 0.0018911362, ...
http://arxiv.org/abs/2306.01303v1
DistilXLSR: A Light Weight Cross-Lingual Speech Representation Model
Haoyu Wang, Siyuan Wang, Wei-Qiang Zhang, Jinfeng Bai
cs.CL, cs.SD, eess.AS
Multilingual self-supervised speech representation models have greatly enhanced the speech recognition performance for low-resource languages, and the compression of these huge models has also become a crucial prerequisite for their industrial application. In this paper, we propose DistilXLSR, a distilled cross-lingual...
http://arxiv.org/pdf/2306.01303v1
2023-06-02T07:03:06
2023-06-02T07:03:06
{ "title": [ -0.0031871796, -0.009605408, -0.038635254, -0.07946777, 0.0063438416, 0.0069007874, -0.08166504, 0.007259369, 0.010055542, 0.046142578, 0.001367569, -0.015312195, 0.020217896, -0.022079468, -0.0463562, 0.020996094, -0.012908936, 0.05...
http://arxiv.org/abs/2306.01296v1
Improved Training for End-to-End Streaming Automatic Speech Recognition Model with Punctuation
Hanbyul Kim, Seunghyun Seo, Lukas Lee, Seolki Baek
eess.AS, cs.CL
Punctuated text prediction is crucial for automatic speech recognition as it enhances readability and impacts downstream natural language processing tasks. In streaming scenarios, the ability to predict punctuation in real-time is particularly desirable but presents a difficult technical challenge. In this work, we pro...
http://arxiv.org/pdf/2306.01296v1
2023-06-02T06:46:14
2023-06-02T06:46:14
{ "title": [ -0.014427185, 0.021728516, -0.048980713, -0.07287598, 0.015777588, -0.0025310516, -0.057617188, 0.014892578, -0.018493652, 0.046417236, -0.01676941, -0.002380371, 0.0067634583, -0.021530151, -0.03717041, -0.0019273758, -0.011177063, ...
http://arxiv.org/abs/2306.03102v1
ChatGPT is a Remarkable Tool -- For Experts
Amos Azaria, Rina Azoulay, Shulamit Reches
cs.HC, cs.AI, cs.CL, cs.CY
This paper investigates the capabilities of ChatGPT as an automated assistant in diverse domains, including scientific writing, mathematics, education, programming, and healthcare. We explore the potential of ChatGPT to enhance productivity, streamline problem-solving processes, and improve writing style. Furthermore, ...
http://arxiv.org/pdf/2306.03102v1
2023-06-02T06:28:21
2023-06-02T06:28:21
{ "title": [ 0.030822754, -0.027374268, 0.0023345947, -0.04095459, -0.033050537, -0.0002799034, -0.009231567, 0.0005311966, -0.021316528, 0.048706055, 0.015701294, 0.025894165, -0.008331299, -0.023483276, 0.001200676, 0.023010254, -0.014678955, 0...
http://arxiv.org/abs/2306.01286v1
KL-Divergence Guided Temperature Sampling
Chung-Ching Chang, David Reitter, Renat Aksitov, Yun-Hsuan Sung
cs.CL, cs.AI
Temperature sampling is a conventional approach to diversify large language model predictions. As temperature increases, the prediction becomes diverse but also vulnerable to hallucinations -- generating tokens that are sensible but not factual. One common approach to mitigate hallucinations is to provide source/ground...
http://arxiv.org/pdf/2306.01286v1
2023-06-02T06:11:26
2023-06-02T06:11:26
{ "title": [ -0.016815186, 0.012145996, -0.04336548, -0.03390503, -0.021392822, -0.001791954, -0.0473938, -0.03250122, -0.008003235, 0.047698975, 0.0023860931, 0.015472412, 0.041015625, -0.04095459, 0.0031414032, 0.033172607, 0.02822876, 0.011222...
http://arxiv.org/abs/2306.01273v1
VoteTRANS: Detecting Adversarial Text without Training by Voting on Hard Labels of Transformations
Hoang-Quoc Nguyen-Son, Seira Hidano, Kazuhide Fukushima, Shinsaku Kiyomoto, Isao Echizen
cs.CL
Adversarial attacks reveal serious flaws in deep learning models. More dangerously, these attacks preserve the original meaning and escape human recognition. Existing methods for detecting these attacks need to be trained using original/adversarial data. In this paper, we propose detection without training by voting on...
http://arxiv.org/pdf/2306.01273v1
2023-06-02T05:18:19
2023-06-02T05:18:19
{ "title": [ -0.025222778, 0.009567261, -0.029052734, -0.009513855, 0.014968872, -0.015556335, 0.0036506653, -0.028900146, -0.017074585, 0.016067505, -0.0035190582, 0.024261475, -0.024658203, -0.021026611, -0.029052734, -0.014099121, -0.011550903, ...
http://arxiv.org/abs/2306.01261v1
Automatic Translation of Hate Speech to Non-hate Speech in Social Media Texts
Yevhen Kostiuk, Atnafu Lambebo Tonja, Grigori Sidorov, Olga Kolesnikova
cs.CL
In this paper, we investigate the issue of hate speech by presenting a novel task of translating hate speech into non-hate speech text while preserving its meaning. As a case study, we use Spanish texts. We provide a dataset and several baselines as a starting point for further research in the task. We evaluated our ba...
http://arxiv.org/pdf/2306.01261v1
2023-06-02T04:03:14
2023-06-02T04:03:14
{ "title": [ 0.00076913834, 0.008155823, -0.026657104, -0.034484863, 0.005252838, -0.015991211, -0.0017490387, -0.011543274, -0.017837524, 0.05758667, -0.020401001, 0.025253296, 0.039642334, -0.0004081726, -0.045837402, 0.010559082, -0.009216309, ...
http://arxiv.org/abs/2306.01248v2
How Ready are Pre-trained Abstractive Models and LLMs for Legal Case Judgement Summarization?
Aniket Deroy, Kripabandhu Ghosh, Saptarshi Ghosh
cs.CL, cs.IR, cs.LG
Automatic summarization of legal case judgements has traditionally been attempted by using extractive summarization methods. However, in recent years, abstractive summarization models are gaining popularity since they can generate more natural and coherent summaries. Legal domain-specific pre-trained abstractive summar...
http://arxiv.org/pdf/2306.01248v2
2023-06-14T07:25:42
2023-06-02T03:16:19
{ "title": [ 0.0017976761, -0.028366089, -0.023376465, -0.0657959, -0.064208984, 0.0134887695, -0.07324219, 0.004081726, 0.031433105, 0.041015625, -0.07336426, 0.0034923553, 0.015396118, -0.0031795502, -0.043548584, -0.023345947, -0.00084781647, ...
http://arxiv.org/abs/2306.01245v1
THiFLY Research at SemEval-2023 Task 7: A Multi-granularity System for CTR-based Textual Entailment and Evidence Retrieval
Yuxuan Zhou, Ziyu Jin, Meiwei Li, Miao Li, Xien Liu, Xinxin You, Ji Wu
cs.CL
The NLI4CT task aims to entail hypotheses based on Clinical Trial Reports (CTRs) and retrieve the corresponding evidence supporting the justification. This task poses a significant challenge, as verifying hypotheses in the NLI4CT task requires the integration of multiple pieces of evidence from one or two CTR(s) and th...
http://arxiv.org/pdf/2306.01245v1
2023-06-02T03:09:31
2023-06-02T03:09:31
{ "title": [ -0.015914917, -0.01828003, -0.04498291, -0.08026123, -0.011039734, 0.028259277, -0.07055664, 0.010612488, 0.009361267, 0.04421997, -0.02053833, 0.047851562, 0.025848389, -0.00066185, -0.031951904, 0.032043457, -0.027069092, 0.0286560...
http://arxiv.org/abs/2306.01242v1
Responsible Task Automation: Empowering Large Language Models as Responsible Task Automators
Zhizheng Zhang, Xiaoyi Zhang, Wenxuan Xie, Yan Lu
cs.AI, cs.CL
The recent success of Large Language Models (LLMs) signifies an impressive stride towards artificial general intelligence. They have shown a promising prospect in automatically completing tasks upon user instructions, functioning as brain-like coordinators. The associated risks will be revealed as we delegate an increa...
http://arxiv.org/pdf/2306.01242v1
2023-06-02T02:42:58
2023-06-02T02:42:58
{ "title": [ -0.016494751, -0.017486572, -0.059265137, -0.10638428, 0.015197754, 0.00029921532, -0.061401367, 0.026123047, -0.004322052, 0.037109375, -0.0011081696, -0.021118164, 0.020965576, -0.025848389, -0.03213501, -0.00006663799, -0.015411377, ...
http://arxiv.org/abs/2306.01807v1
Word Embeddings for Banking Industry
Avnish Patel
cs.CL, cs.AI
Applications of Natural Language Processing (NLP) are plentiful, from sentiment analysis to text classification. Practitioners rely on static word embeddings (e.g. Word2Vec or GloVe) or static word representation from contextual models (e.g. BERT or ELMo) to perform many of these NLP tasks. These widely available word ...
http://arxiv.org/pdf/2306.01807v1
2023-06-02T01:00:44
2023-06-02T01:00:44
{ "title": [ -0.013389587, -0.014144897, -0.03277588, -0.022109985, -0.044891357, -0.021514893, -0.0501709, -0.0016775131, -0.005455017, 0.04272461, -0.06286621, -0.0025997162, 0.031280518, -0.005596161, -0.008308411, -0.009895325, -0.0129776, 0....
http://arxiv.org/abs/2306.01208v2
Adapting an Unadaptable ASR System
Rao Ma, Mengjie Qian, Mark J. F. Gales, Kate M. Knill
eess.AS, cs.CL, cs.SD
As speech recognition model sizes and training data requirements grow, it is increasingly common for systems to only be available via APIs from online service providers rather than having direct access to models themselves. In this scenario it is challenging to adapt systems to a specific target domain. To address this...
http://arxiv.org/pdf/2306.01208v2
2023-10-10T09:44:24
2023-06-01T23:54:11
{ "title": [ 0.026351929, -0.011070251, -0.0072250366, -0.034942627, -0.061462402, -0.0012674332, -0.039398193, -0.012908936, 0.0074501038, 0.051208496, -0.043060303, -0.06402588, 0.011070251, -0.022201538, -0.023712158, -0.01928711, 0.038970947, ...
http://arxiv.org/abs/2306.01206v1
Estimating Semantic Similarity between In-Domain and Out-of-Domain Samples
Rhitabrat Pokharel, Ameeta Agrawal
cs.CL, cs.AI, cs.LG
Prior work typically describes out-of-domain (OOD) or out-of-distribution (OODist) samples as those that originate from dataset(s) or source(s) different from the training set but for the same task. When compared to in-domain (ID) samples, the models have been known to usually perform poorer on OOD samples, although th...
http://arxiv.org/pdf/2306.01206v1
2023-06-01T23:39:07
2023-06-01T23:39:07
{ "title": [ -0.015586853, -0.0063438416, -0.038208008, -0.03643799, 0.028289795, -0.014602661, -0.008407593, -0.028305054, 0.006729126, 0.030166626, -0.023788452, 0.015701294, 0.00223732, -0.012550354, -0.043151855, -0.0020332336, -0.010475159, ...
http://arxiv.org/abs/2306.01201v1
Learning When to Speak: Latency and Quality Trade-offs for Simultaneous Speech-to-Speech Translation with Offline Models
Liam Dugan, Anshul Wadhawan, Kyle Spence, Chris Callison-Burch, Morgan McGuire, Victor Zordan
cs.CL, cs.LG, cs.SD, eess.AS
Recent work in speech-to-speech translation (S2ST) has focused primarily on offline settings, where the full input utterance is available before any output is given. This, however, is not reasonable in many real-world scenarios. In latency-sensitive applications, rather than waiting for the full utterance, translations...
http://arxiv.org/pdf/2306.01201v1
2023-06-01T23:29:23
2023-06-01T23:29:23
{ "title": [ -0.019546509, -0.00086832047, -0.010520935, -0.053009033, -0.014717102, 0.0055656433, -0.04534912, -0.001449585, -0.029647827, 0.042938232, -0.009346008, 0.009552002, 0.0028686523, -0.02230835, -0.035888672, -0.01687622, -0.0021305084, ...
http://arxiv.org/abs/2306.01200v1
Multi-Dimensional Evaluation of Text Summarization with In-Context Learning
Sameer Jain, Vaishakh Keshava, Swarnashree Mysore Sathyendra, Patrick Fernandes, Pengfei Liu, Graham Neubig, Chunting Zhou
cs.CL
Evaluation of natural language generation (NLG) is complex and multi-dimensional. Generated text can be evaluated for fluency, coherence, factuality, or any other dimensions of interest. Most frameworks that perform such multi-dimensional evaluation require training on large manually or synthetically generated datasets...
http://arxiv.org/pdf/2306.01200v1
2023-06-01T23:27:49
2023-06-01T23:27:49
{ "title": [ 0.002828598, 0.011817932, -0.030639648, -0.048034668, 0.004673004, -0.008384705, -0.01852417, -0.012878418, 0.011039734, 0.04751587, -0.031280518, 0.030258179, -0.0027694702, 0.009468079, -0.057159424, -0.015625, -0.013572693, 0.0355...
http://arxiv.org/abs/2306.01183v1
Systematic Evaluation of GPT-3 for Zero-Shot Personality Estimation
Adithya V Ganesan, Yash Kumar Lal, August Håkan Nilsson, H. Andrew Schwartz
cs.CL, 68T50, J.4; I.2; I.7
Very large language models (LLMs) perform extremely well on a spectrum of NLP tasks in a zero-shot setting. However, little is known about their performance on human-level NLP problems which rely on understanding psychological concepts, such as assessing personality traits. In this work, we investigate the zero-shot ab...
http://arxiv.org/pdf/2306.01183v1
2023-06-01T22:43:37
2023-06-01T22:43:37
{ "title": [ 0.0036411285, 0.0069885254, -0.027801514, -0.027130127, 0.0035591125, -0.0206604, -0.01272583, -0.0132369995, -0.003850937, 0.026992798, -0.0010280609, -0.00065755844, 0.0047454834, -0.039794922, -0.022949219, 0.030532837, -0.013755798, ...
http://arxiv.org/abs/2306.01169v1
Hybrid Long Document Summarization using C2F-FAR and ChatGPT: A Practical Study
Guang Lu, Sylvia B. Larcher, Tu Tran
cs.CL
Text summarization is a downstream natural language processing (NLP) task that challenges the understanding and generation capabilities of language models. Considerable progress has been made in automatically summarizing short texts, such as news articles, often leading to satisfactory results. However, summarizing lon...
http://arxiv.org/pdf/2306.01169v1
2023-06-01T21:58:33
2023-06-01T21:58:33
{ "title": [ -0.0045776367, -0.0024757385, -0.0038433075, -0.041870117, -0.02935791, 0.009727478, -0.06542969, -0.01838684, -0.025558472, 0.054901123, -0.0052261353, 0.033325195, 0.008880615, -0.0099105835, -0.033691406, 0.003660202, -0.024749756, ...
http://arxiv.org/abs/2306.01164v1
Leveraging Natural Language Processing For Public Health Screening On YouTube: A COVID-19 Case Study
Ahrar Bin Aslam, Zafi Sherhan Syed, Muhammad Faiz Khan, Asghar Baloch, Muhammad Shehram Shah Syed
cs.CL, cs.SI
Background: Social media platforms have become a viable source of medical information, with patients and healthcare professionals using them to share health-related information and track diseases. Similarly, YouTube, the largest video-sharing platform in the world contains vlogs where individuals talk about their illne...
http://arxiv.org/pdf/2306.01164v1
2023-06-01T21:40:48
2023-06-01T21:40:48
{ "title": [ 0.01209259, 0.0018882751, -0.059783936, -0.08154297, 0.013793945, -0.012580872, -0.030563354, 0.011810303, -0.02420044, 0.014450073, -0.032043457, 0.013130188, -0.012451172, -0.018051147, -0.0138168335, -0.023483276, -0.015914917, 0....
http://arxiv.org/abs/2306.01160v1
Faster Causal Attention Over Large Sequences Through Sparse Flash Attention
Matteo Pagliardini, Daniele Paliotta, Martin Jaggi, François Fleuret
cs.LG, cs.AI, cs.CL
Transformer-based language models have found many diverse applications requiring them to process sequences of increasing length. For these applications, the causal self-attention -- which is the only component scaling quadratically w.r.t. the sequence length -- becomes a central concern. While many works have proposed ...
http://arxiv.org/pdf/2306.01160v1
2023-06-01T21:33:59
2023-06-01T21:33:59
{ "title": [ 0.022415161, 0.010040283, -0.050109863, -0.08325195, 0.03463745, -0.023605347, -0.00548172, 0.0055503845, 0.029296875, 0.041900635, -0.010971069, -0.0011510849, 0.018600464, -0.014045715, -0.043029785, 0.016357422, -0.0006928444, 0.0...
http://arxiv.org/abs/2306.03789v1
On the Robustness of Arabic Speech Dialect Identification
Peter Sullivan, AbdelRahim Elmadany, Muhammad Abdul-Mageed
eess.AS, cs.CL, cs.LG
Arabic dialect identification (ADI) tools are an important part of the large-scale data collection pipelines necessary for training speech recognition models. As these pipelines require application of ADI tools to potentially out-of-domain data, we aim to investigate how vulnerable the tools may be to this domain shift...
http://arxiv.org/pdf/2306.03789v1
2023-06-01T21:31:00
2023-06-01T21:31:00
{ "title": [ 0.0038700104, 0.017105103, -0.04714966, -0.062347412, 0.025970459, -0.011940002, -0.03390503, -0.0032310486, -0.025512695, 0.038482666, 0.014808655, -0.008033752, -0.013725281, -0.03982544, -0.055541992, 0.0017652512, 0.0067710876, 0...
http://arxiv.org/abs/2306.01153v2
Diverse and Faithful Knowledge-Grounded Dialogue Generation via Sequential Posterior Inference
Yan Xu, Deqian Kong, Dehong Xu, Ziwei Ji, Bo Pang, Pascale Fung, Ying Nian Wu
cs.CL
The capability to generate responses with diversity and faithfulness using factual knowledge is paramount for creating a human-like, trustworthy dialogue system. Common strategies either adopt a two-step paradigm, which optimizes knowledge selection and response generation separately, and may overlook the inherent corr...
http://arxiv.org/pdf/2306.01153v2
2023-08-05T05:48:41
2023-06-01T21:23:13
{ "title": [ -0.021011353, -0.002714157, -0.056610107, -0.09637451, 0.0029010773, -0.00016188622, -0.07171631, 0.017913818, 0.00957489, 0.0065727234, 0.008621216, -0.008354187, -0.030181885, -0.038482666, -0.052612305, -0.011650085, -0.011558533, ...
http://arxiv.org/abs/2306.01150v1
Did You Read the Instructions? Rethinking the Effectiveness of Task Definitions in Instruction Learning
Fan Yin, Jesse Vig, Philippe Laban, Shafiq Joty, Caiming Xiong, Chien-Sheng Jason Wu
cs.CL, cs.AI
Large language models (LLMs) have shown impressive performance in following natural language instructions to solve unseen tasks. However, it remains unclear whether models truly understand task definitions and whether the human-written definitions are optimal. In this paper, we systematically study the role of task def...
http://arxiv.org/pdf/2306.01150v1
2023-06-01T21:11:24
2023-06-01T21:11:24
{ "title": [ -0.037841797, -0.0016593933, -0.044708252, -0.014259338, 0.0058135986, -0.038116455, -0.0074157715, -0.033599854, -0.010467529, 0.008720398, -0.018676758, -0.0039596558, -0.024612427, -0.025772095, -0.033355713, -0.01864624, 0.0018720627...
http://arxiv.org/abs/2306.01144v1
Evaluating the Capabilities of Multi-modal Reasoning Models with Synthetic Task Data
Nathan Vaska, Victoria Helus
cs.LG, cs.CL
The impressive advances and applications of large language and joint language-and-visual understanding models has led to an increased need for methods of probing their potential reasoning capabilities. However, the difficulty of gather naturally-occurring data for complex multi-modal reasoning tasks bottlenecks the eva...
http://arxiv.org/pdf/2306.01144v1
2023-06-01T20:56:34
2023-06-01T20:56:34
{ "title": [ -0.015159607, -0.00019705296, -0.048187256, -0.04901123, 0.0028820038, -0.013557434, -0.05569458, -0.016296387, 0.007205963, 0.0128479, -0.020385742, -0.0103302, -0.024215698, -0.024108887, -0.043701172, -0.015792847, -0.01751709, 0....
http://arxiv.org/abs/2306.01128v2
Learning Transformer Programs
Dan Friedman, Alexander Wettig, Danqi Chen
cs.LG, cs.CL
Recent research in mechanistic interpretability has attempted to reverse-engineer Transformer models by carefully inspecting network weights and activations. However, these approaches require considerable manual effort and still fall short of providing complete, faithful descriptions of the underlying algorithms. In th...
http://arxiv.org/pdf/2306.01128v2
2023-10-31T00:47:31
2023-06-01T20:27:01
{ "title": [ 0.0181427, 0.02017212, -0.015556335, -0.006000519, -0.03540039, -0.021194458, -0.02406311, -0.019256592, -0.032165527, 0.032714844, -0.03756714, -0.015007019, 0.006462097, -0.03640747, 0.026626587, -0.022399902, 0.037322998, 0.003576...
http://arxiv.org/abs/2306.01117v1
Examining the Causal Effect of First Names on Language Models: The Case of Social Commonsense Reasoning
Sullam Jeoung, Jana Diesner, Halil Kilicoglu
cs.CL
As language models continue to be integrated into applications of personal and societal relevance, ensuring these models' trustworthiness is crucial, particularly with respect to producing consistent outputs regardless of sensitive attributes. Given that first names may serve as proxies for (intersectional) socio-demog...
http://arxiv.org/pdf/2306.01117v1
2023-06-01T20:05:05
2023-06-01T20:05:05
{ "title": [ -0.03555298, -0.018875122, -0.029418945, -0.045043945, -0.038330078, 0.0054626465, -0.07336426, -0.011741638, -0.0076522827, 0.0206604, -0.002735138, 0.014282227, 0.026382446, -0.024124146, -0.007369995, 0.014442444, -0.033081055, 0....
http://arxiv.org/abs/2306.01116v1
The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Only
Guilherme Penedo, Quentin Malartic, Daniel Hesslow, Ruxandra Cojocaru, Alessandro Cappelli, Hamza Alobeidli, Baptiste Pannier, Ebtesam Almazrouei, Julien Launay
cs.CL, cs.AI
Large language models are commonly trained on a mixture of filtered web data and curated high-quality corpora, such as social media conversations, books, or technical papers. This curation process is believed to be necessary to produce performant models with broad zero-shot generalization abilities. However, as larger ...
http://arxiv.org/pdf/2306.01116v1
2023-06-01T20:03:56
2023-06-01T20:03:56
{ "title": [ -0.0026550293, -0.0036678314, -0.051330566, -0.066223145, 0.0010213852, -0.014602661, -0.06512451, 0.031829834, 0.00032162666, 0.015975952, -0.06732178, 0.028701782, 0.010871887, -0.018356323, 0.019073486, -0.025878906, -0.012817383, ...
http://arxiv.org/abs/2306.01105v2
Revisiting Hate Speech Benchmarks: From Data Curation to System Deployment
Atharva Kulkarni, Sarah Masud, Vikram Goyal, Tanmoy Chakraborty
cs.CL
Social media is awash with hateful content, much of which is often veiled with linguistic and topical diversity. The benchmark datasets used for hate speech detection do not account for such divagation as they are predominantly compiled using hate lexicons. However, capturing hate signals becomes challenging in neutral...
http://arxiv.org/pdf/2306.01105v2
2023-06-15T12:37:34
2023-06-01T19:36:52
{ "title": [ 0.0134887695, 0.0005927086, -0.019714355, -0.050842285, 0.014060974, -0.022445679, 0.004131317, -0.0031871796, -0.019958496, 0.04385376, -0.01737976, 0.034973145, -0.008979797, -0.008232117, -0.05432129, -0.006629944, -0.0005378723, ...
http://arxiv.org/abs/2306.01102v6
LLMatic: Neural Architecture Search via Large Language Models and Quality Diversity Optimization
Muhammad U. Nasir, Sam Earle, Julian Togelius, Steven James, Christopher Cleghorn
cs.NE, cs.AI, cs.CL
Large Language Models (LLMs) have emerged as powerful tools capable of accomplishing a broad spectrum of tasks. Their abilities span numerous areas, and one area where they have made a significant impact is in the domain of code generation. In this context, we view LLMs as mutation and crossover tools. Meanwhile, Quali...
http://arxiv.org/pdf/2306.01102v6
2023-10-04T06:51:09
2023-06-01T19:33:21
{ "title": [ -0.0395813, -0.0032291412, -0.033599854, -0.030151367, 0.011810303, -0.0005455017, -0.045410156, -0.025344849, -0.012084961, 0.034240723, -0.0055618286, -0.00070858, -0.005218506, -0.02684021, -0.028717041, 0.0107421875, -0.007663727, ...
http://arxiv.org/abs/2306.01093v1
UCAS-IIE-NLP at SemEval-2023 Task 12: Enhancing Generalization of Multilingual BERT for Low-resource Sentiment Analysis
Dou Hu, Lingwei Wei, Yaxin Liu, Wei Zhou, Songlin Hu
cs.CL, cs.AI
This paper describes our system designed for SemEval-2023 Task 12: Sentiment analysis for African languages. The challenge faced by this task is the scarcity of labeled data and linguistic resources in low-resource settings. To alleviate these, we propose a generalized multilingual system SACL-XLMR for sentiment analys...
http://arxiv.org/pdf/2306.01093v1
2023-06-01T19:10:09
2023-06-01T19:10:09
{ "title": [ -0.0012226105, -0.026687622, -0.023071289, -0.06365967, -0.0071487427, 0.0049095154, -0.053222656, 0.0025539398, 0.01965332, 0.0647583, -0.02178955, 0.009147644, 0.020324707, 0.011894226, -0.06719971, 0.018554688, -0.026565552, 0.052...
http://arxiv.org/abs/2306.01090v1
Improving the Robustness of Summarization Systems with Dual Augmentation
Xiuying Chen, Guodong Long, Chongyang Tao, Mingzhe Li, Xin Gao, Chengqi Zhang, Xiangliang Zhang
cs.CL
A robust summarization system should be able to capture the gist of the document, regardless of the specific word choices or noise in the input. In this work, we first explore the summarization models' robustness against perturbations including word-level synonym substitution and noise. To create semantic-consistent su...
http://arxiv.org/pdf/2306.01090v1
2023-06-01T19:04:17
2023-06-01T19:04:17
{ "title": [ 0.032196045, 0.0135650635, -0.024505615, -0.09301758, 0.013687134, 0.004211426, -0.053863525, -0.009155273, 0.014221191, 0.054870605, -0.03286743, 0.03717041, -0.006126404, 0.016799927, -0.055908203, 0.010215759, -0.0033035278, 0.020...
http://arxiv.org/abs/2306.01805v1
Cook-Gen: Robust Generative Modeling of Cooking Actions from Recipes
Revathy Venkataramanan, Kaushik Roy, Kanak Raj, Renjith Prasad, Yuxin Zi, Vignesh Narayanan, Amit Sheth
cs.CL, cs.AI, cs.IR
As people become more aware of their food choices, food computation models have become increasingly popular in assisting people in maintaining healthy eating habits. For example, food recommendation systems analyze recipe instructions to assess nutritional contents and provide recipe recommendations. The recent and rem...
http://arxiv.org/pdf/2306.01805v1
2023-06-01T18:49:47
2023-06-01T18:49:47
{ "title": [ -0.0041389465, 0.008605957, -0.050750732, -0.06304932, 0.02078247, -0.017120361, -0.03729248, -0.016738892, 0.0054512024, 0.034973145, 0.014427185, 0.018844604, 0.019195557, -0.035125732, -0.030654907, 0.023895264, -0.0045928955, 0.0...
http://arxiv.org/abs/2306.01076v2
Quantization-Aware and Tensor-Compressed Training of Transformers for Natural Language Understanding
Zi Yang, Samridhi Choudhary, Siegfried Kunzmann, Zheng Zhang
cs.CL, cs.AI
Fine-tuned transformer models have shown superior performances in many natural language tasks. However, the large model size prohibits deploying high-performance transformer models on resource-constrained devices. This paper proposes a quantization-aware tensor-compressed training approach to reduce the model size, ari...
http://arxiv.org/pdf/2306.01076v2
2023-07-08T04:29:09
2023-06-01T18:32:08
{ "title": [ -0.014633179, 0.0036716461, -0.024887085, -0.057922363, 0.005039215, 0.008590698, -0.04144287, -0.0016651154, 0.002023697, 0.03353882, -0.01725769, -0.015090942, 0.012924194, -0.019104004, -0.05758667, -0.00667572, 0.0029125214, 0.02...
http://arxiv.org/abs/2306.01069v1
TimelineQA: A Benchmark for Question Answering over Timelines
Wang-Chiew Tan, Jane Dwivedi-Yu, Yuliang Li, Lambert Mathias, Marzieh Saeidi, Jing Nathan Yan, Alon Y. Halevy
cs.CL, cs.AI, cs.IR
Lifelogs are descriptions of experiences that a person had during their life. Lifelogs are created by fusing data from the multitude of digital services, such as online photos, maps, shopping and content streaming services. Question answering over lifelogs can offer personal assistants a critical resource when they try...
http://arxiv.org/pdf/2306.01069v1
2023-06-01T18:17:13
2023-06-01T18:17:13
{ "title": [ 0.002199173, -0.015136719, -0.047821045, -0.09082031, 0.018981934, -0.0026397705, -0.029251099, 0.019180298, 0.03451538, 0.01928711, -0.014808655, 0.017562866, -0.01626587, -0.0077590942, -0.0625, 0.012413025, -0.004508972, 0.0159301...
http://arxiv.org/abs/2306.01061v1
Reimagining Retrieval Augmented Language Models for Answering Queries
Wang-Chiew Tan, Yuliang Li, Pedro Rodriguez, Richard James, Xi Victoria Lin, Alon Halevy, Scott Yih
cs.CL, cs.DB
We present a reality check on large language models and inspect the promise of retrieval augmented language models in comparison. Such language models are semi-parametric, where models integrate model parameters and knowledge from external data sources to make their predictions, as opposed to the parametric nature of v...
http://arxiv.org/pdf/2306.01061v1
2023-06-01T18:08:51
2023-06-01T18:08:51
{ "title": [ -0.016540527, -0.012359619, -0.039093018, -0.06390381, 0.0029792786, 0.0027332306, -0.04650879, 0.009490967, 0.028030396, 0.047332764, -0.011871338, -0.017044067, 0.019241333, -0.004272461, -0.074645996, 0.017196655, -0.0048828125, 0...
http://arxiv.org/abs/2306.01058v1
Are Layout-Infused Language Models Robust to Layout Distribution Shifts? A Case Study with Scientific Documents
Catherine Chen, Zejiang Shen, Dan Klein, Gabriel Stanovsky, Doug Downey, Kyle Lo
cs.CL
Recent work has shown that infusing layout features into language models (LMs) improves processing of visually-rich documents such as scientific papers. Layout-infused LMs are often evaluated on documents with familiar layout features (e.g., papers from the same publisher), but in practice models encounter documents wi...
http://arxiv.org/pdf/2306.01058v1
2023-06-01T18:01:33
2023-06-01T18:01:33
{ "title": [ -0.07324219, -0.019195557, -0.050811768, -0.026855469, -0.032165527, 0.004245758, -0.081848145, -0.010063171, 0.020202637, 0.039794922, -0.0043792725, -0.0045394897, 0.0052757263, -0.021118164, -0.052246094, 0.011131287, -0.012504578, ...
http://arxiv.org/abs/2306.00976v2
TopEx: Topic-based Explanations for Model Comparison
Shreya Havaldar, Adam Stein, Eric Wong, Lyle Ungar
cs.CL
Meaningfully comparing language models is challenging with current explanation methods. Current explanations are overwhelming for humans due to large vocabularies or incomparable across models. We present TopEx, an explanation method that enables a level playing field for comparing language models via model-agnostic to...
http://arxiv.org/pdf/2306.00976v2
2023-06-02T02:44:03
2023-06-01T17:59:10
{ "title": [ -0.022979736, -0.021881104, -0.027816772, -0.06097412, -0.0056037903, 0.00869751, -0.041748047, -0.030349731, 0.0025501251, 0.0039711, 0.0032806396, 0.034210205, -0.03945923, -0.030181885, -0.004749298, 0.026504517, 0.0015335083, 0.0...
http://arxiv.org/abs/2306.00978v2
AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration
Ji Lin, Jiaming Tang, Haotian Tang, Shang Yang, Xingyu Dang, Chuang Gan, Song Han
cs.CL
Large language models (LLMs) have shown excellent performance on various tasks, but the astronomical model size raises the hardware barrier for serving (memory size) and slows down token generation (memory bandwidth). In this paper, we propose Activation-aware Weight Quantization (AWQ), a hardware-friendly approach for...
http://arxiv.org/pdf/2306.00978v2
2023-10-03T18:20:01
2023-06-01T17:59:10
{ "title": [ -0.06982422, -0.009132385, -0.08648682, -0.03704834, -0.0026416779, -0.01007843, -0.050994873, 0.037506104, -0.024658203, 0.048217773, 0.031677246, 0.013427734, -0.0025959015, -0.04852295, -0.0064964294, 0.015853882, 0.01448822, -0.0...
http://arxiv.org/abs/2306.00947v1
EEL: Efficiently Encoding Lattices for Reranking
Prasann Singhal, Jiacheng Xu, Xi Ye, Greg Durrett
cs.CL, cs.AI, cs.LG
Standard decoding approaches for conditional text generation tasks typically search for an output hypothesis with high model probability, but this may not yield the best hypothesis according to human judgments of quality. Reranking to optimize for "downstream" metrics can better optimize for quality, but many metrics o...
http://arxiv.org/pdf/2306.00947v1
2023-06-01T17:45:32
2023-06-01T17:45:32
{ "title": [ -0.01966858, -0.01210022, -0.019577026, -0.047546387, 0.013084412, 0.015960693, -0.01663208, -0.01638794, 0.02508545, 0.027679443, -0.0027446747, 0.006690979, 0.0118637085, -0.021850586, -0.030654907, 0.03086853, -0.008300781, 0.0075...
http://arxiv.org/abs/2306.00946v2
Exposing Attention Glitches with Flip-Flop Language Modeling
Bingbin Liu, Jordan T. Ash, Surbhi Goel, Akshay Krishnamurthy, Cyril Zhang
cs.LG, cs.CL
Why do large language models sometimes output factual inaccuracies and exhibit erroneous reasoning? The brittleness of these models, particularly when executing long chains of reasoning, currently seems to be an inevitable price to pay for their advanced capabilities of coherently synthesizing knowledge, pragmatics, an...
http://arxiv.org/pdf/2306.00946v2
2023-10-30T16:53:22
2023-06-01T17:44:35
{ "title": [ -0.0075645447, 0.015045166, -0.028289795, -0.036132812, 0.013252258, -0.008804321, -0.021820068, -0.020858765, -0.018112183, 0.035125732, -0.015449524, -0.019104004, 0.029388428, -0.015403748, -0.047180176, -0.00022900105, -0.015396118, ...
http://arxiv.org/abs/2306.00936v2
AMR4NLI: Interpretable and robust NLI measures from semantic graphs
Juri Opitz, Shira Wein, Julius Steen, Anette Frank, Nathan Schneider
cs.CL, cs.IR
The task of natural language inference (NLI) asks whether a given premise (expressed in NL) entails a given NL hypothesis. NLI benchmarks contain human ratings of entailment, but the meaning relationships driving these ratings are not formalized. Can the underlying sentence pair relationships be made more explicit in a...
http://arxiv.org/pdf/2306.00936v2
2023-09-05T13:36:27
2023-06-01T17:39:40
{ "title": [ -0.04031372, -0.0181427, -0.05053711, -0.03466797, -0.011894226, 0.0030765533, -0.049926758, -0.014373779, 0.013694763, 0.0022945404, -0.0262146, 0.014144897, -0.03253174, -0.023452759, -0.034210205, -0.001584053, -0.011062622, 0.037...
http://arxiv.org/abs/2306.00931v1
"Let's not Quote out of Context": Unified Vision-Language Pretraining for Context Assisted Image Captioning
Abisek Rajakumar Kalarani, Pushpak Bhattacharyya, Niyati Chhaya, Sumit Shekhar
cs.CV, cs.CL
Well-formed context aware image captions and tags in enterprise content such as marketing material are critical to ensure their brand presence and content recall. Manual creation and updates to ensure the same is non trivial given the scale and the tedium towards this task. We propose a new unified Vision-Language (VL)...
http://arxiv.org/pdf/2306.00931v1
2023-06-01T17:34:25
2023-06-01T17:34:25
{ "title": [ 0.020629883, -0.0070381165, -0.019302368, -0.05050659, 0.0014715195, -0.034362793, -0.0033016205, -0.009925842, 0.007396698, 0.05407715, -0.027572632, 0.030410767, 0.015701294, 0.0009417534, -0.04156494, 0.0027217865, 0.009101868, 0....
http://arxiv.org/abs/2306.00928v1
ACLM: A Selective-Denoising based Generative Data Augmentation Approach for Low-Resource Complex NER
Sreyan Ghosh, Utkarsh Tyagi, Manan Suri, Sonal Kumar, S Ramaneswaran, Dinesh Manocha
cs.CL, cs.AI, cs.IR
Complex Named Entity Recognition (NER) is the task of detecting linguistically complex named entities in low-context text. In this paper, we present ACLM Attention-map aware keyword selection for Conditional Language Model fine-tuning), a novel data augmentation approach based on conditional generation to address the d...
http://arxiv.org/pdf/2306.00928v1
2023-06-01T17:33:04
2023-06-01T17:33:04
{ "title": [ 0.0046806335, 0.0009675026, -0.047943115, -0.07696533, 0.0024356842, 0.00774765, -0.062561035, -0.003288269, -0.000497818, 0.01197052, -0.017242432, 0.022720337, 0.0041885376, -0.05218506, -0.008720398, 0.011749268, -0.007247925, 0.0...
http://arxiv.org/abs/2306.00924v1
Minding Language Models' (Lack of) Theory of Mind: A Plug-and-Play Multi-Character Belief Tracker
Melanie Sclar, Sachin Kumar, Peter West, Alane Suhr, Yejin Choi, Yulia Tsvetkov
cs.CL, cs.AI, cs.LG
Theory of Mind (ToM)$\unicode{x2014}$the ability to reason about the mental states of other people$\unicode{x2014}$is a key element of our social intelligence. Yet, despite their ever more impressive performance, large-scale neural language models still lack basic theory of mind capabilities out-of-the-box. We posit th...
http://arxiv.org/pdf/2306.00924v1
2023-06-01T17:24:35
2023-06-01T17:24:35
{ "title": [ -0.014198303, -0.021057129, -0.022476196, -0.013496399, -0.032196045, -0.010009766, 0.0039863586, 0.00459671, -0.02029419, 0.019180298, -0.007171631, -0.0071144104, 0.02885437, -0.049072266, 0.012008667, -0.006084442, -0.0065574646, ...
http://arxiv.org/abs/2306.00905v1
T2IAT: Measuring Valence and Stereotypical Biases in Text-to-Image Generation
Jialu Wang, Xinyue Gabby Liu, Zonglin Di, Yang Liu, Xin Eric Wang
cs.CL, cs.AI, cs.CV, I.2.6
Warning: This paper contains several contents that may be toxic, harmful, or offensive. In the last few years, text-to-image generative models have gained remarkable success in generating images with unprecedented quality accompanied by a breakthrough of inference speed. Despite their rapid progress, human biases tha...
http://arxiv.org/pdf/2306.00905v1
2023-06-01T17:02:51
2023-06-01T17:02:51
{ "title": [ 0.006084442, 0.009056091, -0.02192688, -0.0143585205, -0.0047950745, -0.012001038, -0.01210022, -0.04534912, 0.009605408, 0.03579712, -0.02758789, 0.029571533, 0.009979248, -0.01235199, -0.020080566, 0.012062073, -0.017288208, 0.0307...
http://arxiv.org/abs/2306.00890v1
LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day
Chunyuan Li, Cliff Wong, Sheng Zhang, Naoto Usuyama, Haotian Liu, Jianwei Yang, Tristan Naumann, Hoifung Poon, Jianfeng Gao
cs.CV, cs.CL
Conversational generative AI has demonstrated remarkable promise for empowering biomedical practitioners, but current investigations focus on unimodal text. Multimodal conversational AI has seen rapid progress by leveraging billions of image-text pairs from the public web, but such general-domain vision-language models...
http://arxiv.org/pdf/2306.00890v1
2023-06-01T16:50:07
2023-06-01T16:50:07
{ "title": [ -0.026947021, -0.02810669, -0.044830322, -0.030548096, -0.01977539, -0.018325806, -0.027893066, 0.0020389557, -0.025482178, 0.03152466, -0.010475159, -0.0005502701, 0.023086548, -0.043060303, -0.004501343, -0.01864624, 0.01058197, 0....
http://arxiv.org/abs/2306.00887v2
OpenPI-C: A Better Benchmark and Stronger Baseline for Open-Vocabulary State Tracking
Xueqing Wu, Sha Li, Heng Ji
cs.CL
Open-vocabulary state tracking is a more practical version of state tracking that aims to track state changes of entities throughout a process without restricting the state space and entity space. OpenPI is to date the only dataset annotated for open-vocabulary state tracking. However, we identify issues with the datas...
http://arxiv.org/pdf/2306.00887v2
2023-06-20T19:47:20
2023-06-01T16:48:20
{ "title": [ -0.02633667, -0.02204895, -0.026626587, -0.055755615, -0.01914978, 0.0029697418, -0.060180664, -0.030914307, -0.0109939575, 0.0031719208, -0.0440979, 0.030593872, -0.016723633, -0.033813477, -0.0003836155, -0.01902771, 0.0027637482, ...
http://arxiv.org/abs/2306.00864v1
A Transformer-based representation-learning model with unified processing of multimodal input for clinical diagnostics
Hong-Yu Zhou, Yizhou Yu, Chengdi Wang, Shu Zhang, Yuanxu Gao, Jia Pan, Jun Shao, Guangming Lu, Kang Zhang, Weimin Li
cs.CV, cs.CL, cs.LG
During the diagnostic process, clinicians leverage multimodal information, such as chief complaints, medical images, and laboratory-test results. Deep-learning models for aiding diagnosis have yet to meet this requirement. Here we report a Transformer-based representation-learning model as a clinical diagnostic aid tha...
http://arxiv.org/pdf/2306.00864v1
2023-06-01T16:23:47
2023-06-01T16:23:47
{ "title": [ 0.010955811, 0.018417358, -0.042541504, -0.0770874, 0.012565613, -0.015731812, -0.06915283, -0.014030457, -0.0068092346, 0.040161133, -0.0079574585, 0.005859375, 0.012313843, -0.033447266, -0.021575928, -0.0021743774, -0.0056762695, ...
http://arxiv.org/abs/2306.00858v1
Adversarial learning of neural user simulators for dialogue policy optimisation
Simon Keizer, Caroline Dockes, Norbert Braunschweiler, Svetlana Stoyanchev, Rama Doddipatla
cs.CL
Reinforcement learning based dialogue policies are typically trained in interaction with a user simulator. To obtain an effective and robust policy, this simulator should generate user behaviour that is both realistic and varied. Current data-driven simulators are trained to accurately model the user behaviour in a dia...
http://arxiv.org/pdf/2306.00858v1
2023-06-01T16:17:16
2023-06-01T16:17:16
{ "title": [ 0.012718201, 0.0014533997, -0.0124435425, -0.028305054, -0.00029087067, -0.007701874, -0.016616821, -0.0256958, -0.045806885, 0.028961182, 0.013435364, 0.009498596, -0.024047852, -0.033294678, -0.026855469, -0.017089844, -0.023071289, ...
http://arxiv.org/abs/2306.00824v1
Zero and Few-shot Semantic Parsing with Ambiguous Inputs
Elias Stengel-Eskin, Kyle Rawlins, Benjamin Van Durme
cs.CL
Despite the ubiquity of ambiguity in natural language, it is often ignored or deliberately removed in semantic parsing tasks, which generally assume that a given surface form has only one correct logical form. We attempt to address this shortcoming by introducing AmP, a framework, dataset, and challenge for parsing wit...
http://arxiv.org/pdf/2306.00824v1
2023-06-01T15:46:36
2023-06-01T15:46:36
{ "title": [ -0.01612854, -0.007820129, -0.03488159, -0.06689453, -0.010627747, 0.0061454773, -0.054473877, 0.0135650635, 0.0051002502, 0.020187378, -0.026275635, -0.0068244934, -0.019012451, -0.0096206665, -0.049621582, -0.017593384, -0.015975952, ...
http://arxiv.org/abs/2306.00804v3
Adaptive Contextual Biasing for Transducer Based Streaming Speech Recognition
Tianyi Xu, Zhanheng Yang, Kaixun Huang, Pengcheng Guo, Ao Zhang, Biao Li, Changru Chen, Chao Li, Lei Xie
cs.SD, cs.CL, eess.AS
By incorporating additional contextual information, deep biasing methods have emerged as a promising solution for speech recognition of personalized words. However, for real-world voice assistants, always biasing on such personalized words with high prediction scores can significantly degrade the performance of recogni...
http://arxiv.org/pdf/2306.00804v3
2023-08-15T04:36:14
2023-06-01T15:33:30
{ "title": [ 0.01524353, 0.020706177, -0.06112671, -0.07348633, 0.0362854, -0.018478394, -0.03390503, 0.000351429, 0.0033397675, 0.060546875, -0.06112671, 0.00008058548, 0.014770508, 0.0041122437, -0.052368164, -0.014884949, -0.00497818, 0.052612...
http://arxiv.org/abs/2306.00802v2
Birth of a Transformer: A Memory Viewpoint
Alberto Bietti, Vivien Cabannes, Diane Bouchacourt, Herve Jegou, Leon Bottou
stat.ML, cs.CL, cs.LG
Large language models based on transformers have achieved great empirical successes. However, as they are deployed more widely, there is a growing need to better understand their internal mechanisms in order to make them more reliable. These models appear to store vast amounts of knowledge from their training data, and...
http://arxiv.org/pdf/2306.00802v2
2023-11-06T22:51:51
2023-06-01T15:30:33
{ "title": [ 0.03353882, 0.009490967, -0.021957397, -0.042541504, -0.008934021, -0.030929565, -0.017318726, -0.008834839, -0.027297974, 0.024551392, -0.032470703, -0.0039787292, -0.0054855347, -0.032073975, -0.019104004, -0.019897461, 0.029800415, ...
http://arxiv.org/abs/2306.00791v1
Modeling and Analyzing Scorer Preferences in Short-Answer Math Questions
Mengxue Zhang, Neil Heffernan, Andrew Lan
cs.CL, cs.AI
Automated scoring of student responses to open-ended questions, including short-answer questions, has great potential to scale to a large number of responses. Recent approaches for automated scoring rely on supervised learning, i.e., training classifiers or fine-tuning language models on a small number of responses wit...
http://arxiv.org/pdf/2306.00791v1
2023-06-01T15:22:05
2023-06-01T15:22:05
{ "title": [ 0.0043525696, 0.0049476624, -0.013839722, -0.005630493, 0.00087976456, -0.026397705, 0.048034668, -0.01864624, 0.005935669, 0.007789612, -0.03390503, -0.013580322, -0.04663086, -0.012199402, -0.0546875, -0.024276733, -0.01322937, 0.0...
http://arxiv.org/abs/2306.00789v1
Improved Cross-Lingual Transfer Learning For Automatic Speech Translation
Sameer Khurana, Nauman Dawalatabad, Antoine Laurent, Luis Vicente, Pablo Gimeno, Victoria Mingote, James Glass
cs.CL, cs.AI, eess.AS, eess.SP
Research in multilingual speech-to-text translation is topical. Having a single model that supports multiple translation tasks is desirable. The goal of this work it to improve cross-lingual transfer learning in multilingual speech-to-text translation via semantic knowledge distillation. We show that by initializing th...
http://arxiv.org/pdf/2306.00789v1
2023-06-01T15:19:06
2023-06-01T15:19:06
{ "title": [ -0.023605347, -0.0016908646, -0.019317627, -0.059448242, 0.001364708, 0.009765625, -0.044067383, 0.0034313202, -0.0044403076, 0.049591064, -0.0090789795, 0.022109985, 0.02078247, -0.012702942, -0.039978027, 0.016708374, -0.003900528, ...
http://arxiv.org/abs/2306.00784v1
Interpretable Math Word Problem Solution Generation Via Step-by-step Planning
Mengxue Zhang, Zichao Wang, Zhichao Yang, Weiqi Feng, Andrew Lan
cs.CL, cs.AI
Solutions to math word problems (MWPs) with step-by-step explanations are valuable, especially in education, to help students better comprehend problem-solving strategies. Most existing approaches only focus on obtaining the final correct answer. A few recent approaches leverage intermediate solution steps to improve f...
http://arxiv.org/pdf/2306.00784v1
2023-06-01T15:16:18
2023-06-01T15:16:18
{ "title": [ -0.029022217, -0.00074768066, -0.027938843, -0.037231445, -0.013954163, 0.0044898987, -0.039489746, -0.011436462, -0.012870789, 0.0011558533, -0.023910522, 0.018341064, -0.05038452, -0.012512207, -0.014816284, -0.027572632, 0.0076522827,...
http://arxiv.org/abs/2306.00774v1
In-Context Learning User Simulators for Task-Oriented Dialog Systems
Silvia Terragni, Modestas Filipavicius, Nghia Khau, Bruna Guedes, André Manso, Roland Mathis
cs.CL, cs.LG
This paper presents a novel application of large language models in user simulation for task-oriented dialog systems, specifically focusing on an in-context learning approach. By harnessing the power of these models, the proposed approach generates diverse utterances based on user goals and limited dialog examples. Unl...
http://arxiv.org/pdf/2306.00774v1
2023-06-01T15:06:11
2023-06-01T15:06:11
{ "title": [ 0.010414124, -0.0069847107, -0.037475586, -0.057006836, 0.004421234, -0.02281189, -0.012687683, 0.006160736, -0.007369995, 0.06112671, -0.014854431, 0.012840271, 0.016494751, -0.009178162, -0.05029297, -0.006549835, -0.015289307, 0.0...
http://arxiv.org/abs/2306.00765v1
Topic-Guided Sampling For Data-Efficient Multi-Domain Stance Detection
Erik Arakelyan, Arnav Arora, Isabelle Augenstein
cs.CL, cs.AI, cs.IR, stat.CO, stat.ML
Stance Detection is concerned with identifying the attitudes expressed by an author towards a target of interest. This task spans a variety of domains ranging from social media opinion identification to detecting the stance for a legal claim. However, the framing of the task varies within these domains, in terms of the...
http://arxiv.org/pdf/2306.00765v1
2023-06-01T15:00:39
2023-06-01T15:00:39
{ "title": [ -0.002609253, 0.0067481995, -0.03552246, -0.055603027, -0.013999939, -0.019180298, -0.034088135, -0.022155762, 0.040893555, 0.026123047, -0.042236328, 0.022201538, 0.0022792816, -0.001909256, -0.03564453, -0.0016469955, -0.0052490234, ...
http://arxiv.org/abs/2306.01031v1
Bypass Temporal Classification: Weakly Supervised Automatic Speech Recognition with Imperfect Transcripts
Dongji Gao, Matthew Wiesner, Hainan Xu, Leibny Paola Garcia, Daniel Povey, Sanjeev Khudanpur
cs.CL, cs.LG, cs.SD, eess.AS
This paper presents a novel algorithm for building an automatic speech recognition (ASR) model with imperfect training data. Imperfectly transcribed speech is a prevalent issue in human-annotated speech corpora, which degrades the performance of ASR models. To address this problem, we propose Bypass Temporal Classifica...
http://arxiv.org/pdf/2306.01031v1
2023-06-01T14:56:19
2023-06-01T14:56:19
{ "title": [ -0.008514404, -0.000009834766, -0.04458618, -0.083862305, 0.0070114136, -0.0027008057, -0.04446411, 0.008010864, -0.009750366, 0.02355957, -0.049560547, 0.011886597, -0.006587982, -0.0050239563, -0.04534912, -0.026657104, -0.010093689, ...
http://arxiv.org/abs/2306.00755v1
Enhancing the Unified Streaming and Non-streaming Model with Contrastive Learning
Yuting Yang, Yuke Li, Binbin Du
cs.CL, cs.SD, eess.AS
The unified streaming and non-streaming speech recognition model has achieved great success due to its comprehensive capabilities. In this paper, we propose to improve the accuracy of the unified model by bridging the inherent representation gap between the streaming and non-streaming modes with a contrastive objective...
http://arxiv.org/pdf/2306.00755v1
2023-06-01T14:50:39
2023-06-01T14:50:39
{ "title": [ 0.03543091, 0.01966858, -0.019607544, -0.054718018, 0.01600647, -0.015182495, -0.026367188, -0.03994751, 0.018157959, 0.0435791, -0.02684021, 0.014663696, -0.00024342537, -0.031799316, -0.036132812, 0.01083374, -0.011764526, 0.049621...
http://arxiv.org/abs/2306.00751v1
Differentiable Tree Operations Promote Compositional Generalization
Paul Soulos, Edward Hu, Kate McCurdy, Yunmo Chen, Roland Fernandez, Paul Smolensky, Jianfeng Gao
cs.CL, cs.LG
In the context of structure-to-structure transformation tasks, learning sequences of discrete symbolic operations poses significant challenges due to their non-differentiability. To facilitate the learning of these symbolic sequences, we introduce a differentiable tree interpreter that compiles high-level symbolic tree...
http://arxiv.org/pdf/2306.00751v1
2023-06-01T14:46:34
2023-06-01T14:46:34
{ "title": [ -0.014968872, -0.0028457642, -0.023605347, -0.1015625, -0.010734558, 0.014198303, -0.06768799, 0.020858765, -0.022750854, 0.018188477, -0.029190063, 0.022964478, -0.038848877, -0.0256958, -0.005256653, -0.018737793, -0.005290985, 0.0...
http://arxiv.org/abs/2306.00745v2
Column Type Annotation using ChatGPT
Keti Korini, Christian Bizer
cs.CL
Column type annotation is the task of annotating the columns of a relational table with the semantic type of the values contained in each column. Column type annotation is an important pre-processing step for data search and data integration in the context of data lakes. State-of-the-art column type annotation methods ...
http://arxiv.org/pdf/2306.00745v2
2023-07-30T12:07:34
2023-06-01T14:40:52
{ "title": [ -0.0016231537, -0.0060768127, -0.035339355, -0.021621704, -0.050109863, -0.010375977, -0.048736572, -0.0005993843, -0.0134887695, 0.05130005, -0.01133728, 0.0065078735, 0.00047826767, -0.02986145, -0.004558563, -0.007785797, 0.013137817,...
http://arxiv.org/abs/2306.00738v1
ReFACT: Updating Text-to-Image Models by Editing the Text Encoder
Dana Arad, Hadas Orgad, Yonatan Belinkov
cs.CL, cs.CV, 68T50, I.2.7
Text-to-image models are trained on extensive amounts of data, leading them to implicitly encode factual knowledge within their parameters. While some facts are useful, others may be incorrect or become outdated (e.g., the current President of the United States). We introduce ReFACT, a novel approach for editing factua...
http://arxiv.org/pdf/2306.00738v1
2023-06-01T14:32:34
2023-06-01T14:32:34
{ "title": [ -0.033081055, -0.001865387, -0.017456055, -0.01751709, -0.049743652, 0.015975952, -0.071899414, -0.019851685, -0.01159668, 0.06100464, -0.040985107, -0.0047721863, 0.031707764, -0.015396118, -0.029296875, -0.023971558, 0.026229858, 0...
http://arxiv.org/abs/2306.00708v1
Boosting the Performance of Transformer Architectures for Semantic Textual Similarity
Ivan Rep, Vladimir Čeperić
cs.CL, I.2.7
Semantic textual similarity is the task of estimating the similarity between the meaning of two texts. In this paper, we fine-tune transformer architectures for semantic textual similarity on the Semantic Textual Similarity Benchmark by tuning the model partially and then end-to-end. We experiment with BERT, RoBERTa, a...
http://arxiv.org/pdf/2306.00708v1
2023-06-01T14:16:53
2023-06-01T14:16:53
{ "title": [ -0.021972656, 0.008132935, -0.04031372, -0.062042236, 0.030319214, 0.0027866364, -0.032287598, -0.0060424805, 0.010719299, 0.06402588, -0.02809143, 0.018463135, 0.047973633, 0.0044059753, -0.0435791, 0.028518677, -0.0033187866, 0.044...
http://arxiv.org/abs/2306.00697v1
How Generative Spoken Language Modeling Encodes Noisy Speech: Investigation from Phonetics to Syntactics
Joonyong Park, Shinnosuke Takamichi, Tomohiko Nakamura, Kentaro Seki, Detai Xin, Hiroshi Saruwatari
cs.CL, cs.AI, eess.AS
We examine the speech modeling potential of generative spoken language modeling (GSLM), which involves using learned symbols derived from data rather than phonemes for speech analysis and synthesis. Since GSLM facilitates textless spoken language processing, exploring its effectiveness is critical for paving the way fo...
http://arxiv.org/pdf/2306.00697v1
2023-06-01T14:07:19
2023-06-01T14:07:19
{ "title": [ -0.023361206, 0.004173279, -0.034088135, -0.060791016, -0.0022411346, 0.0054244995, -0.05960083, 0.00095176697, 0.0037841797, 0.039886475, -0.018951416, 0.010803223, 0.016036987, -0.022277832, -0.040924072, 0.011024475, -0.005821228, ...
http://arxiv.org/abs/2306.00672v1
Towards Argument-Aware Abstractive Summarization of Long Legal Opinions with Summary Reranking
Mohamed Elaraby, Yang Zhong, Diane Litman
cs.CL
We propose a simple approach for the abstractive summarization of long legal opinions that considers the argument structure of the document. Legal opinions often contain complex and nuanced argumentation, making it challenging to generate a concise summary that accurately captures the main points of the legal opinion. ...
http://arxiv.org/pdf/2306.00672v1
2023-06-01T13:44:45
2023-06-01T13:44:45
{ "title": [ 0.014137268, -0.006111145, -0.0018072128, -0.08782959, -0.015609741, 0.028564453, -0.062805176, -0.011947632, 0.010971069, 0.034851074, -0.03274536, 0.010231018, 0.004386902, -0.0045166016, -0.042175293, -0.0062675476, -0.014770508, ...
http://arxiv.org/abs/2306.00667v1
Predicting the Quality of Revisions in Argumentative Writing
Zhexiong Liu, Diane Litman, Elaine Wang, Lindsay Matsumura, Richard Correnti
cs.CL, cs.AI
The ability to revise in response to feedback is critical to students' writing success. In the case of argument writing in specific, identifying whether an argument revision (AR) is successful or not is a complex problem because AR quality is dependent on the overall content of an argument. For example, adding the same...
http://arxiv.org/pdf/2306.00667v1
2023-06-01T13:39:33
2023-06-01T13:39:33
{ "title": [ -0.016052246, 0.000010371208, -0.025314331, -0.048919678, -0.018661499, 0.0040740967, -0.04425049, -0.014862061, -0.034332275, 0.01197052, -0.018203735, 0.0074501038, -0.018493652, -0.020309448, -0.032440186, -0.027297974, -0.030822754, ...
http://arxiv.org/abs/2306.00665v1
Automatic Glossary of Clinical Terminology: a Large-Scale Dictionary of Biomedical Definitions Generated from Ontological Knowledge
François Remy, Thomas Demeester
cs.CL
Background: More than 400,000 biomedical concepts and some of their relationships are contained in SnomedCT, a comprehensive biomedical ontology. However, their concept names are not always readily interpretable by non-experts, or patients looking at their own electronic health records (EHR). Clear definitions or descr...
http://arxiv.org/pdf/2306.00665v1
2023-06-01T13:37:55
2023-06-01T13:37:55
{ "title": [ -0.026992798, -0.008201599, -0.060546875, -0.070617676, 0.01512146, 0.0033283234, -0.03488159, -0.006175995, -0.002828598, 0.03314209, -0.00053215027, 0.012390137, 0.018066406, -0.022521973, -0.018112183, 0.017822266, -0.0057296753, ...
http://arxiv.org/abs/2306.00660v1
Improving Polish to English Neural Machine Translation with Transfer Learning: Effects of Data Volume and Language Similarity
Juuso Eronen, Michal Ptaszynski, Karol Nowakowski, Zheng Lin Chia, Fumito Masui
cs.CL
This paper investigates the impact of data volume and the use of similar languages on transfer learning in a machine translation task. We find out that having more data generally leads to better performance, as it allows the model to learn more patterns and generalizations from the data. However, related languages can ...
http://arxiv.org/pdf/2306.00660v1
2023-06-01T13:34:21
2023-06-01T13:34:21
{ "title": [ -0.04107666, -0.008178711, -0.02243042, -0.004371643, -0.014663696, -0.002248764, -0.0077667236, -0.011764526, -0.014289856, 0.050872803, -0.014373779, 0.010383606, 0.023620605, -0.017044067, -0.034088135, 0.017715454, -0.008377075, ...
http://arxiv.org/abs/2306.00652v1
Explanation Graph Generation via Generative Pre-training over Synthetic Graphs
Han Cui, Shangzhan Li, Yu Zhang, Qi Shi
cs.CL, cs.AI
The generation of explanation graphs is a significant task that aims to produce explanation graphs in response to user input, revealing the internal reasoning process. This task is challenging due to the significant discrepancy between unstructured user queries and structured explanation graphs. Current research common...
http://arxiv.org/pdf/2306.00652v1
2023-06-01T13:20:22
2023-06-01T13:20:22
{ "title": [ -0.019683838, -0.033935547, -0.026779175, -0.08746338, -0.0418396, 0.014266968, -0.076293945, 0.02229309, -0.004108429, -0.011192322, -0.0010709763, 0.009750366, -0.044799805, -0.0357666, 0.0013790131, -0.014266968, -0.014694214, 0.0...
http://arxiv.org/abs/2306.00645v1
Contextual Distortion Reveals Constituency: Masked Language Models are Implicit Parsers
Jiaxi Li, Wei Lu
cs.CL
Recent advancements in pre-trained language models (PLMs) have demonstrated that these models possess some degree of syntactic awareness. To leverage this knowledge, we propose a novel chart-based method for extracting parse trees from masked language models (LMs) without the need to train separate parsers. Our method ...
http://arxiv.org/pdf/2306.00645v1
2023-06-01T13:10:48
2023-06-01T13:10:48
{ "title": [ -0.024963379, 0.022903442, -0.050720215, -0.049194336, -0.013015747, -0.007534027, -0.076416016, -0.013946533, -0.032073975, 0.018447876, -0.019363403, 0.002500534, 0.022445679, -0.029129028, -0.0063095093, -0.013420105, -0.023132324, ...
http://arxiv.org/abs/2306.00639v2
Being Right for Whose Right Reasons?
Terne Sasha Thorn Jakobsen, Laura Cabello, Anders Søgaard
cs.CL, cs.HC
Explainability methods are used to benchmark the extent to which model predictions align with human rationales i.e., are 'right for the right reasons'. Previous work has failed to acknowledge, however, that what counts as a rationale is sometimes subjective. This paper presents what we think is a first of its kind, a c...
http://arxiv.org/pdf/2306.00639v2
2023-10-13T14:28:03
2023-06-01T13:06:43
{ "title": [ 0.027694702, -0.0072402954, -0.0057373047, -0.049438477, -0.039276123, -0.0036373138, -0.016357422, -0.0017910004, -0.06463623, 0.02659607, -0.0023880005, -0.015960693, -0.045806885, -0.030700684, 0.0067481995, -0.05215454, -0.019454956,...
http://arxiv.org/abs/2306.00622v1
ReviewerGPT? An Exploratory Study on Using Large Language Models for Paper Reviewing
Ryan Liu, Nihar B. Shah
cs.CL, cs.AI, cs.DL
Given the rapid ascent of large language models (LLMs), we study the question: (How) can large language models help in reviewing of scientific papers or proposals? We first conduct some pilot studies where we find that (i) GPT-4 outperforms other LLMs (Bard, Vicuna, Koala, Alpaca, LLaMa, Dolly, OpenAssistant, StableLM)...
http://arxiv.org/pdf/2306.00622v1
2023-06-01T12:45:53
2023-06-01T12:45:53
{ "title": [ -0.03753662, -0.016433716, -0.025497437, -0.023620605, -0.02330017, -0.005290985, -0.035736084, -0.020401001, -0.0035190582, 0.042999268, -0.035369873, 0.01335907, 0.012939453, 0.010673523, -0.05166626, -0.0053100586, -0.009590149, 0...
http://arxiv.org/abs/2306.00618v1
Effective Structured Prompting by Meta-Learning and Representative Verbalizer
Weisen Jiang, Yu Zhang, James T. Kwok
cs.CL, cs.AI, cs.LG
Prompt tuning for pre-trained masked language models (MLM) has shown promising performance in natural language processing tasks with few labeled examples. It tunes a prompt for the downstream task, and a verbalizer is used to bridge the predicted token and label prediction. Due to the limited training data, prompt init...
http://arxiv.org/pdf/2306.00618v1
2023-06-01T12:44:33
2023-06-01T12:44:33
{ "title": [ 0.00012588501, 0.0037460327, -0.012649536, -0.07171631, -0.017089844, 0.014770508, -0.07232666, -0.034820557, -0.022445679, 0.022735596, -0.006965637, 0.02218628, -0.03479004, -0.034851074, -0.021759033, -0.02116394, 0.00831604, -0.0...
http://arxiv.org/abs/2306.03773v1
Some voices are too common: Building fair speech recognition systems using the Common Voice dataset
Lucas Maison, Yannick Estève
eess.AS, cs.CL, cs.LG, cs.SD
Automatic speech recognition (ASR) systems become increasingly efficient thanks to new advances in neural network training like self-supervised learning. However, they are known to be unfair toward certain groups, for instance, people speaking with an accent. In this work, we use the French Common Voice dataset to quan...
http://arxiv.org/pdf/2306.03773v1
2023-06-01T11:42:34
2023-06-01T11:42:34
{ "title": [ -0.000790596, -0.0023841858, -0.06750488, -0.07067871, 0.028076172, -0.025634766, -0.024597168, 0.013076782, -0.012794495, 0.041778564, -0.045074463, 0.010261536, 0.002462387, -0.015235901, -0.050598145, -0.011756897, -0.01852417, 0....
http://arxiv.org/abs/2306.00551v2
Enhancing Programming eTextbooks with ChatGPT Generated Counterfactual-Thinking-Inspired Questions
Arun Balajiee Lekshmi Narayanan, Rully Agus Hendrawan, Venktesh V
cs.CL, cs.HC
Digital textbooks have become an integral part of everyday learning tasks. In this work, we consider the use of digital textbooks for programming classes. Generally, students struggle with utilizing textbooks on programming to the maximum, with a possible reason being that the example programs provided as illustration ...
http://arxiv.org/pdf/2306.00551v2
2023-06-06T16:27:57
2023-06-01T11:14:15
{ "title": [ 0.017562866, -0.02796936, 0.005054474, 0.039886475, -0.054016113, -0.024612427, 0.013717651, -0.031982422, 0.019577026, -0.0013160706, -0.03201294, 0.01890564, -0.02104187, -0.031036377, -0.005622864, -0.015327454, -0.024139404, 0.04...
http://arxiv.org/abs/2306.00550v1
Chain-Of-Thought Prompting Under Streaming Batch: A Case Study
Yuxin Tang
cs.LG, cs.AI, cs.CL
Recently, Large Language Models (LLMs) have demonstrated remarkable capabilities. Chain-of-Thought (CoT) has been proposed as a way of assisting LLMs in performing complex reasoning. However, developing effective prompts can be a challenging and labor-intensive task. Many studies come out of some way to automatically c...
http://arxiv.org/pdf/2306.00550v1
2023-06-01T11:11:39
2023-06-01T11:11:39
{ "title": [ 0.03515625, -0.0037574768, -0.019958496, -0.05783081, -0.0016584396, -0.020889282, -0.009231567, -0.0019779205, -0.0051193237, 0.027191162, -0.04864502, 0.0073051453, 0.009513855, -0.011108398, -0.03262329, -0.015090942, 0.010467529, ...
http://arxiv.org/abs/2306.00539v1
A Call for Standardization and Validation of Text Style Transfer Evaluation
Phil Ostheimer, Mayank Nagda, Marius Kloft, Sophie Fellenz
cs.LG, cs.CL
Text Style Transfer (TST) evaluation is, in practice, inconsistent. Therefore, we conduct a meta-analysis on human and automated TST evaluation and experimentation that thoroughly examines existing literature in the field. The meta-analysis reveals a substantial standardization gap in human and automated evaluation. In...
http://arxiv.org/pdf/2306.00539v1
2023-06-01T10:46:08
2023-06-01T10:46:08
{ "title": [ -0.021896362, 0.00472641, -0.05117798, -0.011314392, -0.01739502, -0.02659607, -0.018737793, -0.009941101, -0.022155762, 0.047729492, -0.0041007996, 0.032806396, 0.027832031, -0.026260376, -0.0069389343, 0.017913818, -0.006916046, 0....
http://arxiv.org/abs/2306.00535v1
The Effects of Input Type and Pronunciation Dictionary Usage in Transfer Learning for Low-Resource Text-to-Speech
Phat Do, Matt Coler, Jelske Dijkstra, Esther Klabbers
cs.CL, eess.AS
We compare phone labels and articulatory features as input for cross-lingual transfer learning in text-to-speech (TTS) for low-resource languages (LRLs). Experiments with FastSpeech 2 and the LRL West Frisian show that using articulatory features outperformed using phone labels in both intelligibility and naturalness. ...
http://arxiv.org/pdf/2306.00535v1
2023-06-01T10:42:56
2023-06-01T10:42:56
{ "title": [ 0.0065994263, 0.0041122437, -0.033050537, -0.03475952, -0.0015182495, -0.016647339, -0.0066375732, 0.0060920715, -0.022476196, 0.048095703, -0.029556274, 0.02305603, 0.0073127747, -0.0084991455, -0.03994751, -0.014808655, -0.020935059, ...
http://arxiv.org/abs/2306.00526v4
Layout and Task Aware Instruction Prompt for Zero-shot Document Image Question Answering
Wenjin Wang, Yunhao Li, Yixin Ou, Yin Zhang
cs.CL, cs.AI, cs.CV
Layout-aware pre-trained models has achieved significant progress on document image question answering. They introduce extra learnable modules into existing language models to capture layout information within document images from text bounding box coordinates obtained by OCR tools. However, extra modules necessitate p...
http://arxiv.org/pdf/2306.00526v4
2023-09-07T08:40:16
2023-06-01T10:28:12
{ "title": [ -0.005634308, -0.030578613, -0.018859863, -0.025756836, -0.0440979, -0.01802063, -0.01991272, 0.020843506, 0.05532837, 0.03753662, -0.04534912, -0.01701355, 0.021240234, -0.0135650635, -0.05859375, -0.00630188, 0.024597168, -0.007205...
http://arxiv.org/abs/2306.00503v1
MEWL: Few-shot multimodal word learning with referential uncertainty
Guangyuan Jiang, Manjie Xu, Shiji Xin, Wei Liang, Yujia Peng, Chi Zhang, Yixin Zhu
cs.CL, cs.AI, cs.CV, cs.LG
Without explicit feedback, humans can rapidly learn the meaning of words. Children can acquire a new word after just a few passive exposures, a process known as fast mapping. This word learning capability is believed to be the most fundamental building block of multimodal understanding and reasoning. Despite recent adv...
http://arxiv.org/pdf/2306.00503v1
2023-06-01T09:54:31
2023-06-01T09:54:31
{ "title": [ -0.033294678, -0.01423645, -0.044708252, -0.011985779, -0.03591919, -0.016647339, -0.024902344, 0.0065841675, 0.01234436, 0.017837524, -0.04537964, 0.03967285, -0.013420105, -0.012161255, -0.022949219, -0.0115737915, -0.008552551, 0....
http://arxiv.org/abs/2306.00502v1
Revisiting Event Argument Extraction: Can EAE Models Learn Better When Being Aware of Event Co-occurrences?
Yuxin He, Jingyue Hu, Buzhou Tang
cs.CL, cs.AI
Event co-occurrences have been proved effective for event extraction (EE) in previous studies, but have not been considered for event argument extraction (EAE) recently. In this paper, we try to fill this gap between EE research and EAE research, by highlighting the question that ``Can EAE models learn better when bein...
http://arxiv.org/pdf/2306.00502v1
2023-06-01T09:53:53
2023-06-01T09:53:53
{ "title": [ 0.031951904, 0.012748718, -0.024810791, -0.07714844, 0.019241333, -0.010375977, -0.048675537, -0.03656006, -0.011161804, 0.027389526, 0.032073975, 0.014389038, 0.018035889, -0.0519104, -0.0064926147, 0.023742676, -0.021469116, 0.0208...
http://arxiv.org/abs/2306.00482v1
Inspecting Spoken Language Understanding from Kids for Basic Math Learning at Home
Eda Okur, Roddy Fuentes Alba, Saurav Sahay, Lama Nachman
cs.CY, cs.CL, cs.SD, eess.AS, math.HO
Enriching the quality of early childhood education with interactive math learning at home systems, empowered by recent advances in conversational AI technologies, is slowly becoming a reality. With this motivation, we implement a multimodal dialogue system to support play-based learning experiences at home, guiding kid...
http://arxiv.org/pdf/2306.00482v1
2023-06-01T09:31:57
2023-06-01T09:31:57
{ "title": [ -0.013465881, -0.028625488, -0.01776123, -0.021865845, -0.032318115, 0.01071167, -0.02003479, 0.018173218, -0.05105591, 0.019195557, -0.04156494, -0.018096924, 0.015609741, -0.019515991, 0.030929565, -0.049621582, 0.020141602, -0.007...
http://arxiv.org/abs/2306.00480v2
Parallel Neurosymbolic Integration with Concordia
Jonathan Feldstein, Modestas Jurčius, Efthymia Tsamoura
cs.AI, cs.CL
Parallel neurosymbolic architectures have been applied effectively in NLP by distilling knowledge from a logic theory into a deep model.However, prior art faces several limitations including supporting restricted forms of logic theories and relying on the assumption of independence between the logic and the deep networ...
http://arxiv.org/pdf/2306.00480v2
2023-06-14T14:25:38
2023-06-01T09:30:24
{ "title": [ -0.02330017, -0.0030670166, -0.008071899, -0.02255249, -0.041931152, 0.020233154, -0.044708252, -0.0335083, -0.0051345825, 0.06427002, 0.014984131, -0.024230957, 0.030258179, -0.04714966, -0.044067383, 0.022155762, 0.007411957, 0.046...
http://arxiv.org/abs/2306.00477v4
Make Pre-trained Model Reversible: From Parameter to Memory Efficient Fine-Tuning
Baohao Liao, Shaomu Tan, Christof Monz
cs.CL, cs.AI, cs.LG
Parameter-efficient fine-tuning (PEFT) of pre-trained language models (PLMs) has emerged as a highly successful approach, with training only a small number of parameters without sacrificing performance and becoming the de-facto learning paradigm with the increasing size of PLMs. However, existing PEFT methods are not m...
http://arxiv.org/pdf/2306.00477v4
2023-10-19T16:03:31
2023-06-01T09:26:17
{ "title": [ 0.0025177002, -0.009521484, -0.0206604, -0.0579834, -0.004589081, -0.011291504, -0.024856567, -0.021987915, -0.014419556, 0.028213501, -0.024490356, -0.012542725, -0.015037537, -0.01576233, -0.053955078, -0.015426636, 0.02217102, 0.0...
http://arxiv.org/abs/2306.00458v2
Exploring Anisotropy and Outliers in Multilingual Language Models for Cross-Lingual Semantic Sentence Similarity
Katharina Hämmerl, Alina Fastowski, Jindřich Libovický, Alexander Fraser
cs.CL
Previous work has shown that the representations output by contextual language models are more anisotropic than static type embeddings, and typically display outlier dimensions. This seems to be true for both monolingual and multilingual models, although much less work has been done on the multilingual context. Why the...
http://arxiv.org/pdf/2306.00458v2
2023-06-07T12:55:49
2023-06-01T09:01:48
{ "title": [ -0.038208008, 0.004623413, -0.012649536, -0.020751953, -0.00856781, 0.017944336, -0.04989624, -0.030731201, 0.01335907, 0.04626465, -0.000082850456, 0.03463745, 0.02116394, -0.013290405, -0.041625977, 0.026885986, -0.020202637, 0.050...
http://arxiv.org/abs/2306.00445v1
A big data approach towards sarcasm detection in Russian
A. A. Gurin, T. M. Sadykov, T. A. Zhukov
cs.CL, cs.AI
We present a set of deterministic algorithms for Russian inflection and automated text synthesis. These algorithms are implemented in a publicly available web-service www.passare.ru. This service provides functions for inflection of single words, word matching and synthesis of grammatically correct Russian text. Select...
http://arxiv.org/pdf/2306.00445v1
2023-06-01T08:34:26
2023-06-01T08:34:26
{ "title": [ -0.0017404556, -0.005569458, -0.022842407, -0.05630493, 0.015037537, -0.0025100708, -0.019424438, -0.003232956, 0.007835388, 0.016723633, -0.038024902, -0.047607422, 0.025421143, -0.026260376, -0.04940796, -0.02331543, -0.020507812, ...
http://arxiv.org/abs/2306.00437v1
Responsibility Perspective Transfer for Italian Femicide News
Gosse Minnema, Huiyuan Lai, Benedetta Muscato, Malvina Nissim
cs.CL
Different ways of linguistically expressing the same real-world event can lead to different perceptions of what happened. Previous work has shown that different descriptions of gender-based violence (GBV) influence the reader's perception of who is to blame for the violence, possibly reinforcing stereotypes which see t...
http://arxiv.org/pdf/2306.00437v1
2023-06-01T08:27:00
2023-06-01T08:27:00
{ "title": [ 0.019836426, -0.03387451, -0.043884277, -0.0073280334, -0.03213501, -0.044158936, 0.016433716, 0.014976501, -0.052825928, 0.049743652, 0.02104187, -0.0030899048, 0.05090332, -0.051940918, 0.009117126, -0.009628296, 0.0063972473, -0.0...
http://arxiv.org/abs/2306.00435v1
How Many Answers Should I Give? An Empirical Study of Multi-Answer Reading Comprehension
Chen Zhang, Jiuheng Lin, Xiao Liu, Yuxuan Lai, Yansong Feng, Dongyan Zhao
cs.CL
The multi-answer phenomenon, where a question may have multiple answers scattered in the document, can be well handled by humans but is challenging enough for machine reading comprehension (MRC) systems. Despite recent progress in multi-answer MRC, there lacks a systematic analysis of how this phenomenon arises and how...
http://arxiv.org/pdf/2306.00435v1
2023-06-01T08:22:21
2023-06-01T08:22:21
{ "title": [ -0.023666382, -0.0053520203, -0.031311035, -0.033203125, -0.0044898987, -0.018188477, -0.018478394, -0.017974854, 0.00015854836, 0.022415161, -0.014091492, 0.020736694, -0.028045654, -0.0038089752, -0.054138184, 0.00032663345, -0.0046386...
http://arxiv.org/abs/2306.00434v1
Divide, Conquer, and Combine: Mixture of Semantic-Independent Experts for Zero-Shot Dialogue State Tracking
Qingyue Wang, Liang Ding, Yanan Cao, Yibing Zhan, Zheng Lin, Shi Wang, Dacheng Tao, Li Guo
cs.CL
Zero-shot transfer learning for Dialogue State Tracking (DST) helps to handle a variety of task-oriented dialogue domains without the cost of collecting in-domain data. Existing works mainly study common data- or model-level augmentation methods to enhance the generalization but fail to effectively decouple the semanti...
http://arxiv.org/pdf/2306.00434v1
2023-06-01T08:21:20
2023-06-01T08:21:20
{ "title": [ 0.010910034, -0.0115356445, -0.03466797, -0.07824707, -0.009262085, -0.018707275, -0.03375244, 0.010284424, 0.011474609, 0.027130127, 0.0066070557, 0.020202637, -0.015182495, -0.023712158, -0.030166626, 0.010108948, -0.0018815994, 0....
http://arxiv.org/abs/2306.00424v1
End-to-end Knowledge Retrieval with Multi-modal Queries
Man Luo, Zhiyuan Fang, Tejas Gokhale, Yezhou Yang, Chitta Baral
cs.CL, cs.CV, cs.IR
We investigate knowledge retrieval with multi-modal queries, i.e. queries containing information split across image and text inputs, a challenging task that differs from previous work on cross-modal retrieval. We curate a new dataset called ReMuQ for benchmarking progress on this task. ReMuQ requires a system to retrie...
http://arxiv.org/pdf/2306.00424v1
2023-06-01T08:04:12
2023-06-01T08:04:12
{ "title": [ -0.016021729, 0.004020691, -0.05886841, -0.082092285, 0.03186035, -0.008636475, -0.035858154, 0.0062217712, 0.023971558, 0.038085938, -0.025512695, -0.0009455681, 0.012466431, -0.009841919, -0.04840088, 0.0042495728, -0.0017328262, 0...
http://arxiv.org/abs/2306.00418v2
Uncertainty-Aware Unlikelihood Learning Improves Generative Aspect Sentiment Quad Prediction
Mengting Hu, Yinhao Bai, Yike Wu, Zhen Zhang, Liqi Zhang, Hang Gao, Shiwan Zhao, Minlie Huang
cs.CL, cs.AI
Recently, aspect sentiment quad prediction has received widespread attention in the field of aspect-based sentiment analysis. Existing studies extract quadruplets via pre-trained generative language models to paraphrase the original sentence into a templated target sequence. However, previous works only focus on what t...
http://arxiv.org/pdf/2306.00418v2
2023-06-03T09:48:36
2023-06-01T07:49:06
{ "title": [ -0.00021481514, 0.005760193, -0.059173584, -0.05984497, 0.000041007996, -0.025497437, -0.027862549, 0.020904541, 0.027053833, 0.025299072, -0.03326416, 0.0015850067, 0.0009455681, -0.024765015, -0.0569458, -0.005569458, -0.008331299, ...
http://arxiv.org/abs/2306.00410v1
Towards hate speech detection in low-resource languages: Comparing ASR to acoustic word embeddings on Wolof and Swahili
Christiaan Jacobs, Nathanaël Carraz Rakotonirina, Everlyn Asiko Chimoto, Bruce A. Bassett, Herman Kamper
cs.CL, cs.SD, eess.AS
We consider hate speech detection through keyword spotting on radio broadcasts. One approach is to build an automatic speech recognition (ASR) system for the target low-resource language. We compare this to using acoustic word embedding (AWE) models that map speech segments to a space where matching words have similar ...
http://arxiv.org/pdf/2306.00410v1
2023-06-01T07:25:10
2023-06-01T07:25:10
{ "title": [ 0.011024475, -0.009918213, 0.00035262108, -0.047546387, -0.017532349, -0.0065078735, -0.036834717, 0.0034542084, -0.019561768, 0.043670654, -0.0034980774, 0.003074646, 0.040222168, -0.033569336, -0.026519775, 0.0075035095, -0.0049476624,...
http://arxiv.org/abs/2306.00400v1
BiSync: A Bilingual Editor for Synchronized Monolingual Texts
Josep Crego, Jitao Xu, François Yvon
cs.CL
In our globalized world, a growing number of situations arise where people are required to communicate in one or several foreign languages. In the case of written communication, users with a good command of a foreign language may find assistance from computer-aided translation (CAT) technologies. These technologies oft...
http://arxiv.org/pdf/2306.00400v1
2023-06-01T07:03:47
2023-06-01T07:03:47
{ "title": [ -0.053009033, -0.030166626, -0.05029297, -0.05883789, -0.021881104, 0.016464233, -0.06585693, 0.021240234, 0.009849548, 0.052215576, -0.0017004013, -0.024002075, 0.04711914, -0.027557373, -0.049438477, 0.000075638294, -0.01889038, 0....
http://arxiv.org/abs/2306.00398v2
Preference-grounded Token-level Guidance for Language Model Fine-tuning
Shentao Yang, Shujian Zhang, Congying Xia, Yihao Feng, Caiming Xiong, Mingyuan Zhou
cs.CL
Aligning language models (LMs) with preferences is an important problem in natural language generation. A key challenge is that preferences are typically provided at the *sequence level* while LM training and generation both occur at the *token level*. There is, therefore, a *granularity mismatch* between the preferenc...
http://arxiv.org/pdf/2306.00398v2
2023-10-09T23:31:44
2023-06-01T07:00:07
{ "title": [ 0.0005993843, -0.013160706, -0.035491943, -0.08117676, -0.013969421, -0.008560181, -0.051086426, 0.021316528, 0.0007033348, 0.029052734, -0.03527832, -0.00028896332, 0.029647827, -0.028579712, -0.028884888, 0.00617218, -0.01838684, 0...
http://arxiv.org/abs/2306.00377v1
Developing and Building Ontologies in Cyber Security
Muhammad Shoaib Farooq, Muhammad Talha Waseem
cs.CR, cs.AI, cs.CL
Cyber Security is one of the most arising disciplines in our modern society. We work on Cybersecurity domain and in this the topic we chose is Cyber Security Ontologies. In this we gather all latest and previous ontologies and compare them on the basis of different analyzing factors to get best of them. Reason to selec...
http://arxiv.org/pdf/2306.00377v1
2023-06-01T06:19:01
2023-06-01T06:19:01
{ "title": [ -0.01890564, 0.0048103333, -0.053222656, -0.0625, 0.02230835, -0.0101623535, -0.021560669, -0.0039100647, -0.05709839, 0.018478394, -0.0101623535, 0.0074157715, -0.016540527, -0.024276733, -0.0046043396, -0.009399414, -0.0124053955, ...
http://arxiv.org/abs/2306.00374v1
CFL: Causally Fair Language Models Through Token-level Attribute Controlled Generation
Rahul Madhavan, Rishabh Garg, Kahini Wadhawan, Sameep Mehta
cs.CL, cs.AI
We propose a method to control the attributes of Language Models (LMs) for the text generation task using Causal Average Treatment Effect (ATE) scores and counterfactual augmentation. We explore this method, in the context of LM detoxification, and propose the Causally Fair Language (CFL) architecture for detoxifying p...
http://arxiv.org/pdf/2306.00374v1
2023-06-01T06:13:51
2023-06-01T06:13:51
{ "title": [ -0.03704834, -0.02961731, -0.053588867, -0.091552734, -0.0055274963, 0.020187378, -0.08135986, 0.029907227, -0.016311646, 0.029037476, -0.011054993, 0.011566162, 0.02848816, -0.027954102, -0.026184082, 0.015197754, -0.038848877, 0.05...
http://arxiv.org/abs/2306.00369v2
Focused Prefix Tuning for Controllable Text Generation
Congda Ma, Tianyu Zhao, Makoto Shing, Kei Sawada, Manabu Okumura
cs.CL
In a controllable text generation dataset, there exist unannotated attributes that could provide irrelevant learning signals to models that use it for training and thus degrade their performance. We propose focused prefix tuning(FPT) to mitigate the problem and to enable the control to focus on the desired attribute. E...
http://arxiv.org/pdf/2306.00369v2
2023-06-10T12:36:48
2023-06-01T06:00:43
{ "title": [ -0.057861328, -0.01914978, -0.02619934, -0.042053223, -0.036376953, 0.025634766, -0.0725708, -0.0041007996, -0.014465332, 0.05178833, 0.024612427, 0.013023376, 0.0012931824, -0.022338867, -0.041412354, 0.013771057, -0.029876709, 0.02...